Deadline Miss Rate Research Articles

Research on parameter optimisation of dynamic priority scheduling algorithm based on improved reinforcement learning

The dynamic priority scheduling algorithm is one of the real-time scheduling algorithms in a power system. However, it ignores the impact of the weight of each index when selecting indicators that affect scheduling performance. There is no definite objective function relation between weight parameters and scheduling performance. Hence, the heuristic algorithm is difficult to optimise the weight parameters. Aiming to solve this problem, a dynamic priority scheduling algorithm based on improved reinforcement learning (RL) is proposed for parameter optimisation. By learning from each other, the weighting parameters and the deadline miss rate (DMR), the global optimisation of weighting parameters can be achieved, but the learning efficiency of the conventional RL method is low. According to the task scheduling performance (the DMR) and the task characteristics, this study improves the RL action step and reward function, which accelerates the online learning speed and improves the optimisation ability of the RL algorithm. Experimental results show that the improved RL algorithm not only optimises the weight parameters but reduces the DMR, which reduces the number of iterations of RL. A scheduling algorithm optimised by RL can be better applied to industrial control and power system resource scheduling, which not only improves control efficiency but reduces scheduling costs.

Solar Power Prediction Assisted Intra-task Scheduling for Nonvolatile Sensor Nodes

With the advent of the era of trillion sensors, solar-powered sensor nodes are widely used as they do not require battery charging or replacement. However, the limited and intermittent solar energy supply seriously affects deadline miss rate (DMR) of tasks. Furthermore, traditional solar-powered sensor nodes also suffer from energy loss of battery charging and voltage conversion. Recently, a storage-less and converter-less power supply architecture has been proposed to achieve higher energy efficiency by removing the leaky energy storage and dc voltage conversion. Without energy storages, a node using inter-task scheduling is more sensitive to solar variations, which results in high DMRs. This paper proposes an intra-task scheduling scheme for the storage-less and converter-less solar-powered sensor nodes, whose features include power prediction based on classified solar profiles, a trigger mechanism to select scheduling points, an artificial neural network to calculate task priorities and a fine-grained task selection algorithm. Experimental results show that the proposed algorithm reduces DMR by up to 30% and improves energy utilization efficiency by 20% with trivial energy overheads.

Co-scheduling Deadline-Sensitive Applications in Large-scale Grid Systems

In large-scale grid systems, plenty of applications are constrained by soft or hard deadline requirement. However, it is difficult to guarantee the deadline requirements of these applications because of the dynamical nature of distributed systems. In this paper, a novel approach is proposed to evaluate the deadline-guarantee of co-allocation schemes that obtained from conventional co-allocation policies. Based on this approach, a hybrid-policy co-allocation model is also proposed to address the issue of deadline-constrained resource co-allocation in grid environments. The proposed model integrates multiple co-allocation policies to generate different co-allocation schemes, and selects the one with optimal deadline-guarantee for grid applications. By this way, the hybrid-policy model combines the merits of different policies, and overcomes the shortcomings of those policies. Extensive simulations are conducted to verify the effectiveness and the performance of the proposed model in terms of deadline-miss rate. Experimental results show that it can provide coallocation scheme with enhanced deadline-guarantee as well as lower deadline-miss rate.

Multi-scheme co-scheduling framework for high-performance real-time applications in heterogeneous grids

Due to unpredictable workload, task co-scheduling with constraint to deadline is a challenging issue in heterogeneous grid environments. To address the problem of scheduling real-time applications, in this paper, a multi-scheme co-scheduling framework MSCSF is proposed aiming to provide enhanced deadline-guarantee in heterogeneous environments. The proposed framework integrates multiple co-scheduling schemes in an extensible manner. By quantitatively evaluating the deadline-guarantee of each co-scheduling scheme, the system can adaptively select the scheme with optimal deadline-guarantee for real-time applications at runtime. In this way, MSCSF combines the merits of existing co-scheduling schemes and overcomes the shortcomings of them. Extensive experiments are conducted to verify the effectiveness and the performance of the proposed framework in terms of deadline-miss rate and mean execution time. Experimental results show that it can provide enhanced deadline-guarantee as well as lower deadline-miss rate for real-time applications in heterogeneous grid environments.

Hybrid-Policy Co-allocation Model in Computational Grid

To address the issue of resource co-allocation with deadline constraint in grids, a novel approach is proposed to evaluate the deadline-guarantee of co-allocation schemes that obtained from conventional co-allocation policies. Based on this approach, a hybrid-policy co-allocation model is also proposed to address the issue of deadline-constrained resource co-allocation in grid environments. The proposed model combines multiple co-allocation policies and selects the one with optimal deadline-guarantee for scheduling. In this way, the hybrid-policy model combines the merits of different policies, and overcomes the shortcomings of those policies. Extensive simulations are conducted to verify the effectiveness and the performance of the proposed model in terms of deadline-miss rate. Experimental results show that it can provide co-allocation scheme with improved deadline-guarantee and lower down the deadline-miss rate for real-time applications.

Workload-aware Reliability Evaluation Model in Grid Computing

To address the issue of reliability evaluation in dynamic grid environment, a workload-aware reliability evaluation model is proposed, in which queuing system is applied to describe the dynamic workload and working of grid resources. Meanwhile, a new type of resource fault, namely Deadline-Miss fault, is introduced to evaluate the reliability of real-time jobs. The validity of the proposed model and its approach to calculate real-time job’s reliability are presented theoretically. Extensive experiments are conducted to verify its performance, and the results show that the proposed model can significantly improve the accuracy of reliability evaluation in presence of dynamic workload. Also, scheduling algorithm based on this model can reduce the jobs’ mean response time and the deadline-miss rate.

A Novel Penalty Controllable Dynamic Voltage Scaling Scheme for Mobile Multimedia Applications

Multimedia decoding is one of the key parts of many popular mobile multimedia applications, such as video telephony, streaming, and video playback. Since the multimedia decoding consumes a significant amount of energy on processors, it is crucial to lower the power consumption and prolong the battery life. In this paper, the statistical analysis of more than 600 processor load trace files is first presented. From the analysis, we found that it is feasible to predict the processor load of multimedia applications accurately using a low order linear time series model if the load is sampled using the feature period, which is obtained with fast Fourier transformation. Based on the analysis, we propose a novel interval-based DVS scheme to achieve penalty controllable energy reduction. The DVS scheme does not need any task profile or involvement of application program, and it is compatible with the service model of general purpose mobile operating systems. In addition, the proposed DVS scheme can handle the nonstationary behavior using an efficient online change detector, and important parameters, such as coefficients of the linear time series model, are estimated on the fly. More importantly, the proposed scheme can keep the overscaling rate (OSR) around a certain predefined value. Since the OSR has a simple and stable relationship with the deadline miss rate (DMR), the penalty incurred by DVS is effectively controlled. Experimental results show that the proposed DVS scheme achieves a much smaller prediction error than previous approaches and achieves a significant processor energy reduction with adjustable and controlled penalty